The Egyptian Journal of Radiology and Nuclear Medicine 47 (2016) 1765–1771

Contents lists available at ScienceDirect

The Egyptian Journal of Radiology and Nuclear Medicine

journal homepage: www.sciencedirect .com/ locate /e j rnm

Original Article

Diffusion MRI versus ultrasound in superficial and deependometriosis

http://dx.doi.org/10.1016/j.ejrnm.2016.07.0110378-603X/� 2016 The Egyptian Society of Radiology and Nuclear Medicine. Production and hosting by Elsevier.This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

⇑ Corresponding author.E-mail address: myaftah@yahoo.com (M.Y. Tadros).

Peer-review under responsibility of The Egyptian Society of Radiology andNuclear Medicine.

Mary Y. Tadros ⇑, Nermeen N. KeriakosRadiology Department, Faculty of Medicine, Ain Shams University, Cairo, Egypt

a r t i c l e i n f o a b s t r a c t

Article history:Received 4 April 2016Accepted 10 July 2016Available online 17 August 2016

Keywords:Magnetic resonance images (MRI)Ultrasound scanning (US)Deep infiltrating endometriosis (DIE)

Background: Endometriosis is a common and clinically important problem in women ofchildbearing age. It is classically defined as the presence of functional endometrial glandsand stroma outside the uterine cavity and musculature. It may vary from microscopicendometriotic implants to large cysts (endometriomas).Objective: To assess the role of ultrasound and magnetic resonance imaging including dif-fusion weighted sequences in superficial and deep endometriosis.Patients and methods: The studied group included 30 patients who were previously clini-cally diagnosed to have endometriotic lesions between October 2013 and November2015. These were sent to our department to identify the extent of lesions and clarify theexact location for proper treatment. All patients were evaluated with ultrasound includingpelvic and Transvaginal and conventional MRI with diffusion weighted images. The sensi-tivity, specificity and diagnostic accuracy for both examinations were calculated.Results: Transabdominal ultrasound examination showed a sensitivity of 81%, specificity of38% with overall accuracy rate of 73% while transvaginal US showed sensitivity of 88%,specificity of 33% and overall accuracy of 76%. By using conventional MRI the sensitivityshowed increase in sensitivity which was 85%, specificity which was 86% and accuracywhich was 85%. By the addition of diffusion weighted MRI sensitivity improved to be97%, specificity 86% and overall accuracy 95%.Conclusion: MRI is the most useful technique for determining the extent of endometriosis,especially in the ultrasonographically-indeterminate suspected masses and deep infiltrat-ing lesions as those of the ureters, bladder, and rectosigmoid.� 2016 The Egyptian Society of Radiology and Nuclear Medicine. Production and hosting byElsevier. This is an open access article under the CC BY-NC-ND license (http://creativecom-

mons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Three primary types of endometriosis are identifiedincluding superficial peritoneal lesions also known asSampson’s syndrome, ovarian endometriomas, and deepinfiltrating endometriosis (DIE) [1].

Solid endometriosis, which is also referred to as deeppelvic endometriosis or deeply infiltrative endometriosis(DIE), is defined by the extension of endometrial glandsand stroma at least 5 mm beneath the peritoneal surface[2]. Unlike endometriomas, which contain viscous pro-teinaceous and hemorrhagic contents, solid masses ofendometriosis are composed of ectopic endometrial glandand stromal cells embedded within dense fibrous tissueand smooth muscle [3]. These lesions are considered veryactive and are strongly associated with pelvic pain symp-toms [4]. DIE classically affects the rectovaginal septum

1766 M.Y. Tadros, N.N. Keriakos / The Egyptian Journal of Radiology and Nuclear Medicine 47 (2016) 1765–1771

and the uterine ligaments (69.2%), the vagina (14.5%), therectosigmoid bowel (9.9%), and the bladder and ureter(6.4%) in the order of frequency [5].

The main manifestations are primary or secondary dys-menorrhea, bleeding disturbances, infertility, dysuria, painon defecation (dyschezia), cycle-dependent or (later) cycle-independent pelvic pain, nonspecific cycle-associated gas-trointestinal or urogenital symptoms [6].

Laparoscopy is the gold standard for the diagnosis ofpelvic endometriosis. However for non-invasive methods,Ultrasound is a readily available and inexpensive tool forthe diagnosis of large endometriosis lesions. Transvaginalultrasound can help diagnose endometriomas, bladderlesions, and deep nodules such as those in the rectovaginalseptum [1].

MRI is helpful in determining the extent of deep infil-trating endometriosis, especially when laparoscopicinspection is limited by adhesions [7]. MRI may help guidesurgical approaches for patients with suspectedendometriosis, especially for deep infiltrating endometrio-sis and other unusual sites of presentation [1].

US scanning and MRI are not sensitive for superficiallesions and in addition that US scanning is not sensitivefor the detection of large implants [8].

Moreover, diffusion-weighted imaging with quantita-tive assessment of apparent diffusion coefficient (ADC) val-ues has been incorporated into pelvic MR imagingprotocols [9]. Endometriomas and solid endometrialimplants demonstrate restricted diffusion [3].

The aim of this study was to assess the additive value ofMRI and DWI over ultrasound in diagnosing endometriosis.

2. Patients and methods

2.1. Patients

The study was done in Ain Shams University Hospital inCairo during the period from October 2013 till November2015. Our study population included 30 women in childbearing period that were previously diagnosed withendometriosis. The patients were referred from Obstetricsand Gynecology Department, Ain Shams University Hospi-tals, for evaluation of extension of the disease. Clinical pre-sentations were as follows: 12 were presented withdysmenorrhea, 2 were presented with infertility, 3 withdyspareunia, and 9 with vague abdominal pain Table 1.Exclusion criteria were patients known to have contraindi-cations for MRI, e.g. an implanted magnetic device andpacemakers.

Table 1The number and percentage of different clinical presentation of 30 womenin child bearing period.

Clinical presentation Number Percentage

Dysmenorrhea 12 40Infertility 2 6.7Dyspareunia 3 10Abdominal pain 9 30Total 30 100

Transvaginal ultrasound and non-contrast MRI weredone in our Radiology department to all patients after sign-ing an informed consent to be enrolled in the study.

All our imaging results were finally compared to thelaparoscopic results which was our gold standard.

2.2. Real time ultrasound technique and image analysis

Transabdominal examination using 3.5 MHZ curvilinearprobe was done with full bladder and the patient lying inthe supine position after good contact gel. Examinationwas held in both longitudinal and transverse sectionsprobing at supra pubic area.

Transvaginal ultrasound probing using 8 MHZ probewas also done in all the patients after complete evacuationof the bladder to confirm our findings and search for addi-tional ones not clearly seen by the Transabdominaltechnique.

During each sonographic examination, lesions identi-fied were as follows:

1. Adnexal cysts with low level homogenous internalechoes representing chocolate cysts.

2. Cysts with heterogeneous contents, retracted clot orwith fishnet appearance representing hemorrhagiccysts.

3. Dilated tubes with low level internal echoes suspectedto be hematosalpinx.

4. Cysts with soft tissue components representing com-plex cysts.

5. Ill-defined hypo echoic lesions representing endometri-otic implants in the following sites: cul de sac, aroundurinary bladder wall and along the visualized bowelloops as well as anterior abdominal wall, usually atthe site of previous surgical scar.

2.3. Acquisition and processing of MRI with and withoutdiffusion

A 1.5 T MR imaging unit (Philips, Healthcare) was usedwith a high resolution body coil.

Sequences obtained were axial and sagittal with andwithout fat-suppressed fast spin-echo T1-weightedsequences: axial, coronal, and sagittal T2-weightedsequences. Cuts were taken obliquely in orientation tothe uterus. Parameters included were FOV 200–250 mm,slice thickness 4 mm, section gap 0.5–1 mm, and matrix192 � 256.

DWI and ADC maps were also done. The diffusionweighted imaging ‘‘DWI”: Axial DWI with single shot echoplanar imaging (EPI) performed at b values of 0 and 800 s/mm2. Contrast material was injected in one patient afterchecking renal functions (according to clinician’s request,to exclude metastatic deposits in this patient with historyof breast cancer) and showed enhancement of solid lesions.

2.4. MRI analysis

The images taken were reviewed by two consultantradiologists for the following:

Table 2Diagnostic criteria of endometriosis by Transabdominal and transvaginalUS.

Findings ABD US TVUS

Unilateral chocolate cyst 8 8Bilateral chocolate cysts. 10 11Hematosalpinx 5 4Ill-defined lesions 4 6Complex ovarian cystic lesions 2 2Hemorrhagic ovarian cysts 2 3No abnormality 3 3

M.Y. Tadros, N.N. Keriakos / The Egyptian Journal of Radiology and Nuclear Medicine 47 (2016) 1765–1771 1767

1. Ovarian cystic lesions with high signal on T1-weightedimages with shading on T2 weighted images represent-ing endometriomas.

2. Dilated fallopian tubes with high signal intensity in T1weighted images, confirming hematosalpinx.

3. Bloody foci seen as high signal lesions on fat suppressedT1-weighted sequences.

4. Fibrotic and deep infiltrating lesions seen as low tointermediate signal intensity lesions on high resolutionT2-weighted images in cul de sac, uterosacral ligament,posterior fornix bladder wall and anterior abdominalwall. These appear as hypo intense masses with irregu-lar, indistinct, or stellate margins or even as irregularand predominantly hypo intense soft-tissue thickeningwith T2-weighted sequences.

5. Diffusion restricted lesions on diffusion weightedimages and ADC map to detect deep infiltrating lesionsnot detected on conventional sequences.

2.5. Statistical analysis

Statistical analysis was done using MedCelc� version12.5 (MedCalc� Software bvba, Ostend, Belgium) andDAG stat (Mackinnon, 2000). To determine the diagnosticvalue of conventional MRI, series of 2 � 2 contingencytables were constructed. The following measures werethen calculated: sensitivity, specificity, and overall diag-nostic accuracy.

3. Results

Our study included 30 women in child bearing periodthat were previously diagnosed with endometriosis rang-ing from 25 to 40 years with the mean age of 32 ± SD.

By Transabdominal ultrasound, 4 cases showed isoe-choic soft tissue lesions, either related to a surgical scaror deep in the pelvis, 5 cases showed dilated tubular struc-tures with internal echoes, 10 showed bilateral adnexalcysts with homogeneous floating internal echoes (choco-late cysts), 8 cases showed unilateral chocolate cysts, 2showed adnexal cysts with hyper echoic components, 2showed complex cysts with soft tissue components, and3 did not show any abnormality.

By using complementary transvaginal ultrasound, con-firmation of the diagnosis was done in 4 out of 5 cases ofhematosalpinx, 11 cases turned to have bilateral ovarianchocolate cysts and 8 out of 8 cases of unilateral ovarianchocolate cysts, 2 out of 2 hemorrhagic adnexal cysts and2 complex cysts with soft tissue components. 1 extra hem-orrhagic adnexal cyst was identified. 6 cases showed suspi-cions ill-defined hypo echoic implants in various positions(Table 2).

As regards the 3 cases of hemorrhagic adnexal cysts,follow-up after 6–8 weeks showed complete disappear-ance of the lesions, so laparoscopic correlation was notneeded.

Total findings exceeded the number of patients andmore than one patient had double pathology picture.

By conventional MRI, 8 out of 11 cases were confirmedto have bilateral chocolate cysts (Fig. 1), while 3 turned out

to be non-hemorrhagic on MRI. Unilateral chocolate cystswere also confirmed by MRI in 7 cases (1 of which wasassociated with enhancing soft tissue components). The 2complex cystic lesions proved by MRI to be hemorrhagiccysts with clots resembling soft tissue on ultrasound. MRIalso confirmed 4 cases of hematosalpinx (Fig. 2), 3 surgicalscar implants and 5 cases of DIE (Figs. 3–5).

After the addition of diffusion weighted sequences, thefollowing results were obtained and compared to laparo-scopic findings (Table 3).

3 surgical scar implants were confirmed, 2 were discov-ered, and 5 DIEs were confirmed.

By that we conclude that Transabdominal ultrasoundexamination showed a sensitivity of 81%, specificity of38% with overall accuracy rate of 73% while transvaginalUS shows sensitivity of 88%, specificity of 33% and overallaccuracy of 76%.

By using conventional MRI the sensitivity showedincrease in sensitivity which was 85%, specificity of 86%and accuracy of 85%. By the addition of diffusion weightedMRI sensitivity improved to be 97%, specificity 86% andoverall accuracy 95%.

So if we compare between US and MRI with adding dif-fusion we will have the following results as listed in(Table 4, Fig. 6).

4. Discussion

Endometriosis, which is defined as the presence of ecto-pic endometrial glands and stroma outside the uterus, is acommon cause of pelvic pain and infertility, affecting asmany as 10% of premenopausal women [3]. The prevalenceof endometriosis is significantly higher in women who areinfertile than in women who are fertile [6].

In our study, 30 patients previously diagnosed to haveendometriosis were presented with different symptoms.Our aim was to confirm or exclude the diagnosis, detectdeep infiltrating lesions and define the extent of thedisease.

Transabdominal ultrasound is usually the first investi-gation done in subject suspected of any pelvic disease. UShas the advantage of having good resolution, easy accessi-bility, less expensive, and is free of ionizing radiation [5].However TVS has superior image quality and resolutionas compared to TAS. The typical ultrasound findingsinclude a cystic mass with diffuse, low level echoes [5].

In our study, we could detect 18 cases of endometri-omas by Transabdominal ultrasound. These were seen as

Fig. 1. (a–d) Left ovarian cystic lesion with bloody component displaying high signal on T1 (a) with shading on T2 (b). It shows restricted diffusion (c) withlow ADC value on ADC map (d). Findings of endometrioma. Right ovarian cyst with peripheral high signal on T1 (a) representing hemorrhagic cyst.

Fig. 2. (a–e) Bilateral hematosalpinx showing high signal on T1 (a) with shading on T2 (b: axial, c: sagittal). These show restricted diffusion (d) with lowADC value on ADC map (e). Hematosalpinx is diagnostic of endometriosis.

1768 M.Y. Tadros, N.N. Keriakos / The Egyptian Journal of Radiology and Nuclear Medicine 47 (2016) 1765–1771

Fig. 3. (a–c) Douglas pouch deep infiltrating endometriotic implant detected by DWI (a) and confirmed by ADC map (b). This was not detected on T2weighted images(c).

Fig. 4. (a–b) Anterior wall bladder endometriotic implant detected as low signal lesion on T2 weighted axial (a) and sagittal (b) images.

Fig. 5. (a–d) Anterior abdominal wall endometriotic implant detected on ultrasound (a) as ill-defined hypo echoic lesion. This was confirmed by MRI as lowsignal lesion on T1 (b) and T2 (c) weighted images with heterogeneous enhancement on post contrast study (d).

M.Y. Tadros, N.N. Keriakos / The Egyptian Journal of Radiology and Nuclear Medicine 47 (2016) 1765–1771 1769

Table 3Correlation of MRI and MRI DW results with laparoscopic results.

Findings MRI MRIDW

Positivelaparoscopic results

Unilateral chocolate cyst 7 8 8Bilateral chocolate cysts 10 11 10Hematosalpinx 4 4 4Ill-defined lesions

(superficial and DIE)8 10 10

Complex ovarian cysticlesions

2 2 0

Hemorrhagic ovarian cysts 3 3 0No abnormality 2 1 1

Table 4Comparison between the sensitivity, specificity and accuracy rate resultsbetween different studies we used (p < 0.05).

Points ofcomparison

US(%)

TVUS(%)

MRI(%)

MRI with diffusion(%)

Sensitivity 81 88 84 97Specificity 38 33 86 86Overall accuracy 73 76 85 76

%0

20%

40%

60%

80%

100%

120%

abd US TV US MRI MRI DW

sensitvity specificity overall accurecy

Fig. 6. Diagrammatic drawing showing the results of our study regardingsensitivity, specificity and overall accuracy %.

1770 M.Y. Tadros, N.N. Keriakos / The Egyptian Journal of Radiology and Nuclear Medicine 47 (2016) 1765–1771

cystic adnexal lesions. 11 appeared with floating fine inter-nal echoes. This was in agreement with Kapoor et al. whosaid that the typical US scan finding in endometriosis is acyst containing low-level homogenous internal echoesconsistent with old blood [10].

In Bagaria’s et al. study they stated that at times, anendometrioma may resemble a cystic-solid or entirelysolid mass [5]. This was seen in our study in 2 casesdetected as complex adnexal lesions. By transvaginal ultra-sound 1 case of bilateral chocolate cysts was discoveredthat was missed by transabdominal ultrasound. This wasin agreement with Bagaria et al. who proved thattransvaginal ultrasound has high sensitivity (92%) andspecificity (99%) in detecting endometriomas, comparedto transabdominal ultrasound.

A study done by Asch and Levine stated that only 60% ofendometriomas larger than 2 cm were appropriately diag-nosed prospectively at sonography. This was due to thevariable sonographic appearance of endometriomas [11].

Moreover MRI could detect a case of unilateral choco-late cyst that was thought to be simple cyst by ultrasound.This showed high signal on T1 weighted images with shad-ing on T2 weighted images. In 1991, Togashi and col-leagues [12] showed that findings of an adnexal masswith high signal intensity on T1-weighted MR imagesand signal intensity lower than that of simple fluid onT2-weighted images helped establish a diagnosis ofendometrioma with specificity greater than 90%.

MRI confirmed our transvaginal sonographic findings of11 cases of bilateral endometriomas. We based our diagno-sis on a study done by Siegelman and Oliver [3] who foundthat bilaterally multifocality of adnexal lesions, along withthe other characteristics can help establish a diagnosis ofendometrioma with even greater specificity than T1 signalhyperintensity alone. However on laparoscopy, 1 of themshowed to be hemorrhagic cysts giving false positiveresults on MRI as this was reported by Sunita el al. [13].

DWIs also helped us to identify and confirm the pres-ence of endometriomas. These showed restricted diffusionwith low ADC value when evaluated in a diffusion-weighted image obtained with a high b value. This was800 in our study. This comes in agreement with Siegelmanand Oliver who stated that endometriomas have less signalintensity to lose on images obtained with higher b valuesthan adnexal masses with higher T2 signal intensity andthat endometriomas often have low ADC values [3].

According to Outwater et al. [14] the presence of T1-weighted hyperintensity within a dilated fallopian tube issuggestive of endometriosis and may be the only findingat MR imaging in some women. Five cases were suspectedto have hematosalpinx by transabdominal ultrasound, oneof which was false positive on transvaginal ultrasound andthe other 4 were confirmed on MRI. Laparoscopic resultswere the same as MRI, giving 100% specificity for MRI incases of hematosalpinx.

MRI is superior to ultrasound in detecting DIE. Thesewere seen as low signal lesions on T2 weighted imagesand confirmed by DWIs. 4 cases were suspected by trans-abdominal ultrasound. 6 were seen also on transvaginalultrasound. On conventional MRI, 8 cases were detected.By adding DWIs with low b value (800), extra 2 cases weredetected. All of which were later confirmed bylaparoscopy.

Our main differential diagnosis with endometriomaswas dermoid and hemorrhagic cysts. Dermoid cysts char-acteristically contain fat which could be easily identifiedon ultrasound as hyperechoic areas with no posterioracoustic shadowing. On MRI they show high signal on T1,which is differentiated by signal dropout on fat suppres-sion images, compared to endometrioma that does notsuppress on T1 fat suppressed images. This was in agree-ment with Natalie Yang et al., who stated that endometri-omas have homogenous high signal intensity on T1 whichdoes not suppress on T1FS, unlike a dermoid which has sig-nal dropout on fat suppression images and chemical shiftartifact [15].

Diffusion MRI is good in differentiation betweenendometriomas and hemorrhagic cysts as endometriomasshow restricted diffusion with low ADC value compared

M.Y. Tadros, N.N. Keriakos / The Egyptian Journal of Radiology and Nuclear Medicine 47 (2016) 1765–1771 1771

to hemorrhagic cysts which usually show facilitateddiffusion.

Other differential diagnosis included ovarian mucinoustumors due to their increased signal on T1, yet being lessintense than fat or blood and do not show diffusion restric-tion [13].

Patel et al. [16] stated that an adnexal mass with diffuselow-level internal echoes and absence of particular neo-plastic features is highly likely to be an endometrioma.US neoplastic features included heterogenous lesions withirregular borders and shape on ultrasound with soft tissuecomponents that show diffusion restriction with very lowADC values and post contrast enhancement on MRI.

5. Conclusion

Ultrasound could detect the adnexal endometrioticlesion; however, MRI has a better specificity in the detec-tion of the nature of these lesions with better specificity.Moreover MRI especially with the use of additional DWIyields better diagnostic accuracy with better detectionand evaluation of the extent of DIE and in confirming thediagnosis.

Conflict of interest

All authors have no conflict of interest.

References

[1] Hsu Albert L, Khachikyan Izabella, Stratton Pamela. Invasive andnon-invasive methods for the diagnosis of endometriosis. Clin ObstetGynecol 2010;53(2):413–9.

[2] Coccia ME, Rizzello F. Ultrasonographic staging: a new stagingsystem for deep endometriosis. Ann N Y Acad Sci 2011;1221:61–9.

[3] Siegelman Evan S, Oliver Edward R. MR imaging of endometriosis:ten imaging pearls. RSNA; Radiographics 2012;32(6).

[4] Koninckx PR, Meuleman C, Demeyere S, Lesaffre E, Cornillie FJ.Suggestive evidence that pelvic endometriosis is a progressivedisease, whereas deeply infiltrating endometriosis is associatedwith pelvic pain. Fertil Steril 1991;55:759–65.

[5] Bagaria Shalini Jain, Rasalkar Darshana D, Paunipagar Bhawan K.Imaging tools for endometriosis: role of ultrasound, MRI and otherimaging modalities in diagnosis and planning intervention. In:Endometriosis – basic concepts and current research trends.Available from: www.intechopen.com.

[6] Halis Gülden, Mechsner Sylvia, Ebert Andreas D. The diagnosis andtreatment of deep infiltrating endometriosis. Dtsch Arztebl Int2010;107(25):446–56.

[7] van Waesberghe Jan Hein, Hazewinkel Mariek, Busard Milou. MRIdetection of endometriosis. Radiol Assist 2015(November 1).Available from: http://www.radiologyassistant.nl/en/p4da490c32edcc/mri-detection-of-endometriosis.html.

[8] Daly Shawn, Outwater Eric K. Imaging in endometrioma/endometriosis. Medscape 2015. Available from: http://emedicine.medscape.com/article/403435-overview.

[9] Coutinho Jr AC, Krishnaraj A, Pires CE, Bittencourt LK, Guimarães AR.Pelvic applications of diffusion magnetic resonance images. MagnReson Imag Clin N Am 2011;19(1):133–57.

[10] Kapoor Dharmesh, Alderman Elizabeth, Davila Willy, Hiraoka MarkKY, et al. Endometriosis workup; 2015. http://emedicine.medscape.com/article/271899-workup.

[11] Asch E, Levine D. Variations in appearance of endometriomas. JUltrasound Med 2007;26(8):993–1002.

[12] Togashi K, Nishimura K, Kimura I, et al. Endometrial cysts: diagnosiswith MR imaging. Radiology 1991;180(1):73–8.

[13] Dhanda Sunita, Thakur Meenakshi, Kerkar Rajendra, JagmohanPooja. Diffusion-weighted imaging of gynecologic tumors:diagnostic pearls and potential pitfalls. Radiographics 2014;34(5):1393–416.

[14] Outwater EK, Siegelman ES, Chiowanich P, Kilger AM, Dunton CJ,Talerman A. Dilated fallopian tubes: MR imaging characteristics.Radiology 1998;208(2):463–9.

[15] Yang Natalie et al. http://radiopaedia.org/articles/endometriosis[last seen June 2016].

[16] Patel Maitray D, Feldstein Vickie A, Chen Dillon C, Lipson Scott D,Filly Roy A. Endometriomas: diagnostic performance of US.Radiology 1999;210(3):739–45.